普通规则匹配 以及规则修改接口

docus-client-interface/src/main/java/com/docus/server/api/segmentation/OcrRuleTestApi.java

@@ -70,6 +70,14 @@ public interface OcrRuleTestApi {
     @PutMapping("/edit")
     void edit(@RequestBody EditOcrRuleTestDTO editOcrRuleTestDTO);
 
+    /**
+     * 删除后保存
+     *
+     * @param editOcrRuleTestDTOList 编辑参数
+     */
+    @PutMapping("/deleteAndAdd")
+    void deleteAndAdd(@RequestBody List<EditOcrRuleTestDTO> editOcrRuleTestDTOList);
+
     /**
      * 批量删除
      *

docus-client-interface/src/main/java/com/docus/server/dto/segmentation/ocrfileinfo/AddOcrFileInfoDTO.java

@@ -67,19 +67,19 @@ public class AddOcrFileInfoDTO implements Serializable {
     private OcrStatusEnum ocrStatus;
 
     @ApiModelProperty(value = "分段id")
-    private Integer assortId;
+    private String assortId;
 
     @ApiModelProperty(value = "分段名称")
     private String assortName;
 
     @ApiModelProperty(value = "命中策略id")
-    private Integer ruleId;
+    private Long ruleId;
 
     @ApiModelProperty(value = "命中关键词")
     private String hitKey;
 
     @ApiModelProperty(value = "命中占比")
-    private Float rate;
+    private Double rate;
 
     @ApiModelProperty(value = "创建时间")
     private Date createTime;

docus-client-interface/src/main/java/com/docus/server/dto/segmentation/ocrfileinfo/EditOcrFileInfoDTO.java

@@ -64,19 +64,19 @@ public class EditOcrFileInfoDTO implements Serializable {
     private OcrStatusEnum ocrStatus;
 
     @ApiModelProperty(value = "分段id")
-    private Integer assortId;
+    private String assortId;
 
     @ApiModelProperty(value = "分段名称")
     private String assortName;
 
     @ApiModelProperty(value = "命中策略id")
-    private Integer ruleId;
+    private Long ruleId;
 
     @ApiModelProperty(value = "命中关键词")
     private String hitKey;
 
     @ApiModelProperty(value = "命中占比")
-    private Float rate;
+    private Double rate;
 
     @ApiModelProperty(value = "创建时间")
     private Date createTime;

docus-client-interface/src/main/java/com/docus/server/dto/segmentation/ocrfileinfotest/AddOcrFileInfoTestDTO.java

@@ -67,19 +67,19 @@ public class AddOcrFileInfoTestDTO implements Serializable {
     private OcrStatusEnum ocrStatus;
 
     @ApiModelProperty(value = "分段id")
-    private Integer assortId;
+    private String assortId;
 
     @ApiModelProperty(value = "分段名称")
     private String assortName;
 
     @ApiModelProperty(value = "命中策略id")
-    private Integer ruleId;
+    private Long ruleId;
 
     @ApiModelProperty(value = "命中关键词")
     private String hitKey;
 
     @ApiModelProperty(value = "命中占比")
-    private Float rate;
+    private Double rate;
 
     @ApiModelProperty(value = "创建时间")
     private Date createTime;

docus-client-interface/src/main/java/com/docus/server/dto/segmentation/ocrfileinfotest/EditOcrFileInfoTestDTO.java

@@ -65,19 +65,19 @@ public class EditOcrFileInfoTestDTO implements Serializable {
     private OcrStatusEnum ocrStatus;
 
     @ApiModelProperty(value = "分段id")
-    private Integer assortId;
+    private String assortId;
 
     @ApiModelProperty(value = "分段名称")
     private String assortName;
 
     @ApiModelProperty(value = "命中策略id")
-    private Integer ruleId;
+    private Long ruleId;
 
     @ApiModelProperty(value = "命中关键词")
     private String hitKey;
 
     @ApiModelProperty(value = "命中占比")
-    private Float rate;
+    private Double rate;
 
     @ApiModelProperty(value = "创建时间")
     private Date createTime;

docus-client-interface/src/main/java/com/docus/server/entity/segmentation/OcrFileInfo.java

@@ -89,7 +89,7 @@ public class OcrFileInfo implements Serializable {
 
     @ApiModelProperty(value = "分段id")
     @TableField("assort_id")
-    private Integer assortId;
+    private String assortId;
 
     @ApiModelProperty(value = "分段名称")
     @TableField("assort_name")
@@ -97,7 +97,7 @@ public class OcrFileInfo implements Serializable {
 
     @ApiModelProperty(value = "命中策略id")
     @TableField("rule_id")
-    private Integer ruleId;
+    private Long ruleId;
 
     @ApiModelProperty(value = "命中关键词")
     @TableField("hit_key")
@@ -105,7 +105,7 @@ public class OcrFileInfo implements Serializable {
 
     @ApiModelProperty(value = "命中占比")
     @TableField("rate")
-    private Float rate;
+    private Double rate;
 
     @ApiModelProperty(value = "文件大小")
     @TableField("file_size")

docus-client-interface/src/main/java/com/docus/server/entity/segmentation/OcrFileInfoTest.java

@@ -90,7 +90,7 @@ public class OcrFileInfoTest implements Serializable {
 
     @ApiModelProperty(value = "分段id")
     @TableField("assort_id")
-    private Integer assortId;
+    private String assortId;
 
     @ApiModelProperty(value = "分段名称")
     @TableField("assort_name")
@@ -98,7 +98,7 @@ public class OcrFileInfoTest implements Serializable {
 
     @ApiModelProperty(value = "命中策略id")
     @TableField("rule_id")
-    private Integer ruleId;
+    private Long ruleId;
 
     @ApiModelProperty(value = "命中关键词")
     @TableField("hit_key")
@@ -106,7 +106,7 @@ public class OcrFileInfoTest implements Serializable {
 
     @ApiModelProperty(value = "命中占比")
     @TableField("rate")
-    private Float rate;
+    private Double rate;
 
     @ApiModelProperty(value = "文件大小")
     @TableField("file_size")

docus-client-interface/src/main/java/com/docus/server/entity/segmentation/OcrRuleJson.java

@@ -17,7 +17,7 @@ public class OcrRuleJson {
     private String assortName;
 
     @ApiModelProperty(value = "占比分")
-    private Integer matchRatio;
+    private Double matchRatio;
 
     @ApiModelProperty(value = "匹配关键词 多个#分割")
     private String keyWord;

docus-client-interface/src/main/java/com/docus/server/entity/segmentation/OcrRuleResult.java

@@ -6,8 +6,18 @@ import lombok.Data;
 @Data
 public class OcrRuleResult {
 
-    @ApiModelProperty(value = "id")
+    @ApiModelProperty(value = "分段id")
-    private Long id;
+    private String assortId;
 
+    @ApiModelProperty(value = "分段名称")
+    private String assortName;
 
+    @ApiModelProperty(value = "命中策略id")
+    private Long ruleId;
+
+    @ApiModelProperty(value = "命中关键词")
+    private String hitKey;
+
+    @ApiModelProperty(value = "命中占比")
+    private Double rate;
 }

docus-client-interface/src/main/java/com/docus/server/entity/segmentation/OcrSpecialResult.java

@@ -7,7 +7,20 @@ import lombok.Data;
 @Data
 public class OcrSpecialResult {
 
-    @ApiModelProperty(value = "id")
+    @ApiModelProperty(value = "分段id")
-    private Long id;
+    private String assortId;
+
+    @ApiModelProperty(value = "分段名称")
+    private String assortName;
+
+    @ApiModelProperty(value = "命中策略id")
+    private Long ruleId;
+
+    @ApiModelProperty(value = "命中关键词")
+    private String hitKey;
+
+    @ApiModelProperty(value = "命中占比")
+    private Double rate;
+
 
 }

docus-client-interface/src/main/java/com/docus/server/entity/segmentation/OcrSpecialRuleJson.java

@@ -11,7 +11,7 @@ import lombok.*;
 public class OcrSpecialRuleJson {
 
     @ApiModelProperty(value = "命中分数")
-    private Integer score;
+    private double score;
 
     @ApiModelProperty(value = "匹配关键词 多个#分割")
     private String keyWord;

docus-client-interface/src/main/java/com/docus/server/enums/OcrStatusEnum.java

@@ -11,7 +11,7 @@ public enum OcrStatusEnum implements IIntegerEnum {
 
     STARTING(1, "分段中"),
 
-    COMPLETE(1, "完成分段");
+    COMPLETE(2, "完成分段");
 
     private Integer value;
     private String display;

docus-client-interface/src/main/java/com/docus/server/vo/segmentation/ocrfileinfo/OcrFileInfoVO.java

@@ -64,19 +64,19 @@ public class OcrFileInfoVO implements Serializable {
     private OcrStatusEnum ocrStatus;
 
     @ApiModelProperty(value = "分段id")
-    private Integer assortId;
+    private String assortId;
 
     @ApiModelProperty(value = "分段名称")
     private String assortName;
 
     @ApiModelProperty(value = "命中策略id")
-    private Integer ruleId;
+    private Long ruleId;
 
     @ApiModelProperty(value = "命中关键词")
     private String hitKey;
 
     @ApiModelProperty(value = "命中占比")
-    private Float rate;
+    private Double rate;
 
     @ApiModelProperty(value = "创建时间")
     private Date createTime;

docus-client-interface/src/main/java/com/docus/server/vo/segmentation/ocrfileinfotest/OcrFileInfoTestVO.java

@@ -64,19 +64,19 @@ public class OcrFileInfoTestVO implements Serializable {
     private OcrStatusEnum ocrStatus;
 
     @ApiModelProperty(value = "分段id")
-    private Integer assortId;
+    private String assortId;
 
     @ApiModelProperty(value = "分段名称")
     private String assortName;
 
     @ApiModelProperty(value = "命中策略id")
-    private Integer ruleId;
+    private Long ruleId;
 
     @ApiModelProperty(value = "命中关键词")
     private String hitKey;
 
     @ApiModelProperty(value = "命中占比")
-    private Float rate;
+    private Double rate;
 
     @ApiModelProperty(value = "创建时间")
     private Date createTime;

docus-segmentation/src/main/java/com/docus/server/controller/OcrRuleTestController.java

@@ -84,6 +84,11 @@ public class OcrRuleTestController implements OcrRuleTestApi {
         iOcrRuleTestService.edit(editOcrRuleTestDTO);
     }
 
+    @Override
+    public void deleteAndAdd(List<EditOcrRuleTestDTO> editOcrRuleTestDTOList) {
+        iOcrRuleTestService.deleteAndAdd(editOcrRuleTestDTOList);
+    }
+
     /**
      * 批量删除
      *

docus-segmentation/src/main/java/com/docus/server/service/IOcrRuleTestService.java

@@ -20,10 +20,10 @@ public interface IOcrRuleTestService {
 
      /**
      * 判断是否命中分段规则
-     * @param text
+     * @param ocrTextList
      * @return
      */
-     OcrRuleResult handleSpecialRule(String text);
+     OcrRuleResult handleRule(List<String> ocrTextList);
 
     /**
      * 按主键查询
@@ -71,4 +71,6 @@ public interface IOcrRuleTestService {
      * @return 分页列表
      */
     PageResult<OcrRuleTestVO> search(SearchDTO searchDTO);
+
+    void deleteAndAdd(List<EditOcrRuleTestDTO> editOcrRuleTestDTOList);
 }

docus-segmentation/src/main/java/com/docus/server/service/IOcrSpecialRuleTestService.java

@@ -31,7 +31,7 @@ public interface IOcrSpecialRuleTestService {
      * @param result
      * @return
      */
-    boolean handleSpecialRule(OcrSpecialResult result,String text);
+    boolean handleSpecialRuleEnd (OcrSpecialResult result,List<String> ocrTextList);
 
 
     /**

docus-segmentation/src/main/java/com/docus/server/service/handler/StartOcrTestHandler.java

@@ -1,14 +1,19 @@
 package com.docus.server.service.handler;
 
+import com.docus.core.util.Func;
 import com.docus.log.annotation.TrackRetryListener;
 import com.docus.server.api.ocr.OcrApi;
 import com.docus.server.entity.segmentation.OcrFileInfoTest;
+import com.docus.server.entity.segmentation.OcrRuleResult;
 import com.docus.server.entity.segmentation.OcrSpecialResult;
+import com.docus.server.enums.OcrStatusEnum;
 import com.docus.server.infrastructure.dao.IOcrFileInfoTestDao;
+import com.docus.server.service.IOcrRuleTestService;
 import com.docus.server.service.IOcrSpecialRuleTestService;
 import org.springframework.stereotype.Component;
 
 import javax.annotation.Resource;
+import java.util.Date;
 import java.util.List;
 
 /**
@@ -22,25 +27,60 @@ public class StartOcrTestHandler {
     private IOcrFileInfoTestDao iOcrFileInfoTestDao;
     @Resource
     private IOcrSpecialRuleTestService iOcrSpecialRuleTestService;
+    @Resource
+    private IOcrRuleTestService iOcrRuleTestService;
+
 
     @TrackRetryListener("START_OCR_TEST")
     public void startOcr(String patientId) {
+        boolean isSpecialFile = false;
+        OcrSpecialResult lastSpecialResul = null;
+
         //根据病案号查询文件列表
         List<OcrFileInfoTest> fileInfoTestList = iOcrFileInfoTestDao.findBy("patientId",patientId);
-
         for (OcrFileInfoTest fileInfoTest:fileInfoTestList) {
+
             //遍历文件列表 获取ocr识别结果
             List<String> ocrTextList = ocrApi.getText(fileInfoTest.getPicCutUrl());
+
+            fileInfoTest.setOcrText(Func.toJson(ocrTextList));
+            fileInfoTest.setOcrStatus(OcrStatusEnum.COMPLETE);
+            fileInfoTest.setOcrFinishTime(new Date());
+            //判断上次是否特殊文件开始  需要判断是否特殊文件结尾
+            if (isSpecialFile){
+                fileInfoTest.setAssortId(lastSpecialResul.getAssortId());
+                fileInfoTest.setAssortName(lastSpecialResul.getAssortName());
+                fileInfoTest.setRuleId(lastSpecialResul.getRuleId());
+                boolean b = iOcrSpecialRuleTestService.handleSpecialRuleEnd(lastSpecialResul, ocrTextList);
+                if (b){
+                    //当前分段为特殊分段结尾
+                    isSpecialFile = false;
+                    lastSpecialResul = null;
+                }
+                continue;
+            }
             //优先判断是否特殊规则开始
             OcrSpecialResult ocrSpecialResult = iOcrSpecialRuleTestService.handleSpecialRule(ocrTextList);
             //判断是否属于某个分段
+            if (ocrSpecialResult!=null){
+                fileInfoTest.setAssortId(ocrSpecialResult.getAssortId());
+                fileInfoTest.setAssortName(ocrSpecialResult.getAssortName());
+                fileInfoTest.setRuleId(ocrSpecialResult.getRuleId());
 
+                isSpecialFile = true;
+                lastSpecialResul = ocrSpecialResult;
+                continue;
+            }
+            OcrRuleResult ocrRuleResult = iOcrRuleTestService.handleRule(ocrTextList);
             //更新文件数据的分段
-
+            if (ocrRuleResult!=null){
+                fileInfoTest.setAssortId(ocrRuleResult.getAssortId());
+                fileInfoTest.setAssortName(ocrRuleResult.getAssortName());
+                fileInfoTest.setRuleId(ocrRuleResult.getRuleId());
+            }
         }
-
         //更新数据库表
-
+        iOcrFileInfoTestDao.updateBatchById(fileInfoTestList,1000);
 
     }
 }

docus-segmentation/src/main/java/com/docus/server/service/impl/OcrRuleTestServiceImpl.java

@@ -1,5 +1,6 @@
 package com.docus.server.service.impl;
 
+import com.docus.core.util.Func;
 import com.docus.infrastructure.redis.service.IdService;
 import com.docus.infrastructure.web.request.SearchDTO;
 import com.docus.infrastructure.web.response.PageResult;
@@ -7,8 +8,7 @@ import com.docus.server.convert.OcrRuleTestConvert;
 import com.docus.server.dto.segmentation.ocrruletest.AddOcrRuleTestDTO;
 import com.docus.server.dto.segmentation.ocrruletest.DeleteOcrRuleTestDTO;
 import com.docus.server.dto.segmentation.ocrruletest.EditOcrRuleTestDTO;
-import com.docus.server.entity.segmentation.OcrRuleResult;
+import com.docus.server.entity.segmentation.*;
-import com.docus.server.entity.segmentation.OcrRuleTest;
 import com.docus.server.infrastructure.dao.IOcrRuleTestDao;
 import com.docus.server.service.IOcrRuleTestService;
 import com.docus.server.vo.segmentation.ocrruletest.OcrRuleTestVO;
@@ -16,7 +16,9 @@ import org.springframework.stereotype.Service;
 import org.springframework.transaction.annotation.Transactional;
 
 import javax.annotation.Resource;
+import java.util.Comparator;
 import java.util.List;
+import java.util.stream.Collectors;
 
 /**
  * 匹配策略表-调试 服务实现类
@@ -32,17 +34,56 @@ public class OcrRuleTestServiceImpl implements IOcrRuleTestService {
     private IdService idService;
 
     @Override
-    public OcrRuleResult handleSpecialRule(String text) {
+    public OcrRuleResult handleRule(List<String> ocrTextList) {
-
         //查询所有规则列表
-
+        List<OcrRuleTest> ocrRuleTestList = iOcrRuleTestDao.findAll();
-
+        //判断是否符合某个条件的开始条件
-        //判断是否包含
+        for (OcrRuleTest ocrRuleTest : ocrRuleTestList) {
-
+            String assortId = ocrRuleTest.getAssortId();
-
+            double score = ocrRuleTest.getMatchRatio();
-        //判断命中率
+            String[] keyWordList = ocrRuleTest.getKeyWord().split("#");
-
+
-
+            for (String ocrText : ocrTextList) {
+                for (int i = 0; i < keyWordList.length; i++) {
+                    String keyWord = keyWordList[i];
+                    if (ocrText.contains(keyWord)) {
+                        double rate = keyWord.length() / (double) ocrText.length() * 100;
+                        //判断是否命中
+                        if (rate >= score) {
+                            //判断二级是否启用
+                            if (ocrRuleTest.getIsUseSecond() && Func.isNotEmpty(ocrRuleTest.getJson())) {
+                                List<OcrRuleJson> ocrRuleList = Func.parseJsonArray(ocrRuleTest.getJson(), OcrRuleJson.class)
+                                        .stream().sorted(Comparator.comparing(OcrRuleJson::getSort))
+                                        .collect(Collectors.toList());
+                                for (OcrRuleJson ruleJson : ocrRuleList) {
+                                    assortId = ruleJson.getAssortId();
+                                    keyWordList = ruleJson.getKeyWord().split("#");
+                                    score = ruleJson.getMatchRatio();
+                                    for (int j = 0; j < keyWordList.length; j++) {
+                                        keyWord = keyWordList[i];
+                                        if (ocrText.contains(keyWord)) {
+                                             rate = keyWord.length() / (double) ocrText.length() * 100;
+                                            //判断是否命中
+                                            if (rate >= score) {
+                                                OcrRuleResult ocrRuleResult = new OcrRuleResult();
+                                                ocrRuleResult.setRuleId(ocrRuleTest.getId());
+                                                ocrRuleResult.setAssortId(assortId);
+                                                return ocrRuleResult;
+                                            }
+                                        }
+                                    }
+                                }
+                            } else {
+                                OcrRuleResult ocrRuleResult = new OcrRuleResult();
+                                ocrRuleResult.setRuleId(ocrRuleTest.getId());
+                                ocrRuleResult.setAssortId(assortId);
+                                return ocrRuleResult;
+                            }
+                        }
+                    }
+                }
+            }
+        }
         return null;
     }
 
@@ -95,6 +136,19 @@ public class OcrRuleTestServiceImpl implements IOcrRuleTestService {
         //return result;
     }
 
+    @Override
+    public void deleteAndAdd(List<EditOcrRuleTestDTO> editOcrRuleTestDTOList) {
+        List<OcrRuleTest> ruleTestList = editOcrRuleTestDTOList.stream().map(m->{
+            OcrRuleTest ocrRuleTest = OcrRuleTestConvert.INSTANCE.convertDO(m);
+            ocrRuleTest.setId(ocrRuleTest.getId()==null?idService.getDateSeq():ocrRuleTest.getId());
+            return ocrRuleTest;
+        }).collect(Collectors.toList());
+        String assortId = ruleTestList.get(0).getAssortId();
+
+        iOcrRuleTestDao.deleteBy("assortId",assortId);
+        iOcrRuleTestDao.saveBatch(ruleTestList,1000);
+    }
+
     /**
      * 新增
      *

docus-segmentation/src/main/java/com/docus/server/service/impl/OcrSpecialRuleTestServiceImpl.java

@@ -1,5 +1,7 @@
 package com.docus.server.service.impl;
 
+import com.docus.core.util.Func;
+import com.docus.core.util.jackson.JsonUtil;
 import com.docus.infrastructure.redis.service.IdService;
 import com.docus.infrastructure.web.request.SearchDTO;
 import com.docus.infrastructure.web.response.PageResult;
@@ -8,10 +10,13 @@ import com.docus.server.dto.segmentation.ocrspecialruletest.AddOcrSpecialRuleTes
 import com.docus.server.dto.segmentation.ocrspecialruletest.DeleteOcrSpecialRuleTestDTO;
 import com.docus.server.dto.segmentation.ocrspecialruletest.EditOcrSpecialRuleTestDTO;
 import com.docus.server.entity.segmentation.OcrSpecialResult;
+import com.docus.server.entity.segmentation.OcrSpecialRule;
+import com.docus.server.entity.segmentation.OcrSpecialRuleJson;
 import com.docus.server.entity.segmentation.OcrSpecialRuleTest;
 import com.docus.server.infrastructure.dao.IOcrSpecialRuleTestDao;
 import com.docus.server.service.IOcrSpecialRuleTestService;
 import com.docus.server.vo.segmentation.ocrspecialruletest.OcrSpecialRuleTestVO;
+import io.swagger.annotations.ApiModelProperty;
 import org.springframework.stereotype.Service;
 import org.springframework.transaction.annotation.Transactional;
 
@@ -31,20 +36,48 @@ public class OcrSpecialRuleTestServiceImpl implements IOcrSpecialRuleTestService
     @Resource
     private IdService idService;
 
+
     @Override
     public OcrSpecialResult handleSpecialRule(List<String> ocrTextList) {
-        // 查询所以的特殊规则列表
-
 
+        // 查询所以的特殊规则列表
+        List<OcrSpecialRuleTest> all = iOcrSpecialRuleTestDao.findAll();
         //判断是否符合某个条件的开始条件
-
+        for (OcrSpecialRuleTest ocrSpecialRuleTest : all) {
-
+
+            String assortId = ocrSpecialRuleTest.getAssortId();
+            String startJson = ocrSpecialRuleTest.getStartJson();
+            String assortName = ocrSpecialRuleTest.getName();
+
+            List<OcrSpecialRuleJson> ocrSpecialRuleJsons = Func.parseJsonArray(startJson, OcrSpecialRuleJson.class);
+            for (OcrSpecialRuleJson ocrSpecialRuleJson : ocrSpecialRuleJsons) {
+                double score = ocrSpecialRuleJson.getScore();
+                String[] keyWordList = ocrSpecialRuleJson.getKeyWord().split("#");
+                for (String ocrText : ocrTextList) {
+                    for (int i = 0; i < keyWordList.length; i++) {
+                        String keyWord = keyWordList[i];
+                        if (ocrText.contains(keyWord)){
+                            double rate = keyWord.length() / (double)ocrText.length() * 100;
+                            //判断是否命中
+                            if (rate>=score) {
+                                OcrSpecialResult ocrSpecialResult = new OcrSpecialResult();
+                                ocrSpecialResult.setAssortId(assortId);
+                                ocrSpecialResult.setAssortName(assortName);
+                                ocrSpecialResult.setRate(rate);
+                                ocrSpecialResult.setHitKey(keyWord);
+                               return ocrSpecialResult;
+                            }
+                        }
+                    }
+                }
+            }
+        }
         //默认返回空对象
         return null;
     }
 
     @Override
-    public boolean handleSpecialRule(OcrSpecialResult result,String text) {
+    public boolean handleSpecialRuleEnd(OcrSpecialResult result, List<String> ocrTextList) {
 
         //获取结束条件列表